Speech output processing
2.2.1 Speech Processing and STM
Some researchers argue for a close relationship between speech processing skills and STM. Following a review o f research findings into the development o f phonological skills, Snowling and Hulme (1994) argue that “we might consider STM mechanisms to be no more than a by-product o f the m echanism s that exist primarily for the perception and production o f speech” (p. 23). Hulme et al., (1991) propose a model whereby verbal rehearsal, considered integral to STM in most current research, is seen as a by-product o f speech production and perception processes. This view is partly supported by Gathercole and Martin (1996) who hypothesise that “the capacity for temporary storage o f verbal material develops as a by product o f the speech input processing system” (p. 73).
The model proposed by Gathercole and Martin (1996) highlights the role o f speech processing in recall. In contrast to the reliance on speech production processes required for sub-vocal rehearsal in the traditional WM model, Gathercole and Martin (1996) emphasise the role o f input skills, and suggest that “there is no separate temporary memory system” (p. 77). Rather, they argue that the perception o f the verbal material to be remembered triggers temporary activation o f phonological representations, and that these representations are then used in recall. This model allows an explanation o f lexicality effects found in STM task performance. These have included better recall for familiar words than for non-words (Hulme,
Maughan & Brown, 1991), for familiar words than for less familiar words (Hulme et al., 1994), and for more ‘word-like’ nonwords than less ‘word like’ nonwords (Gathercole, Willis, Emslie & Baddeley, 1991). These findings will be discussed in more detail in Section 2.4.iii o f this chapter, and in Chapter 4. This ‘interactive’ model (Gathercole & Martin, 1996) suggests that STM performance is reliant on the mechanisms and knowledge used in processing speech, with particular emphasis on speech input mechanisms, and ‘top-down’ processing.
Henry and Millar (1993) also use experimental findings o f effects o f word familiarity to argue for the role o f long-term memory in developmental increases in STM span. These and other findings lead Henry and Millar (1993) to propose a model for the development o f STM span that “has three main constituents : 1) phonological input storage which decays over time; 2) a speech output system which can be used for verbal recall or for active rehearsal o f the material; and 3) long-term lexical memory as well as long term memory for speech output procedures” (p. 278). This model (Henry & Millar, 1993) fits the philosophy o f the current study in which the integrity o f the child’s whole speech processing system is considered to play a determining role in STM performance.
Hulme and Roodenrys (1995) suggest that “performance on a STM test will depend on many different psychological processes, most obviously attentional, perceptual and response mechanisms as well as storage mechanisms” (p. 375). The perceptual and response mechanisms o f STM tests are often speech based. Thus, the contribution o f speech processing to STM performance at input, storage and response stages is acknowledged. So, whilst STM capacity itself may be related to speech processing skill.
performance on STM tasks will also be affected by the speech processing demands o f the task.
Different task paradigms have been used in research into, and assessment of, STM. These tasks often require elements o f input and output speech processing skills and access to phonological and / or semantic representations. Recall tasks may involve participants in listening to spoken material and either repeating it or pointing to pictures that represent the items presented. The spoken material will consist o f words, which may be very familiar or less familiar, or non-words. Spoken presentation will require input skills o f speech perception, discrimination and recognition. If words the listener is familiar with are used, existing phonological and semantic representations are likely to be involved in processing the material and carrying out the task. The presentation o f non-words will have different processing demands. The absence o f existing representations will mean that lexical support is not available to support discrimination o f the speech signal neither will established motor programs be available for the spoken response. Where spoken recall o f the material presented is required, speech output processing skills will affect performance. Again, the use of words or non-words, and, hence, the availability o f existing representations will also affect a listener’s performance.
Other STM tasks may require the presentation o f visual material with either spoken responses or picture pointing responses. Recognition o f the visual material will access semantic representations and if material is to be verbally coded for rehearsal within the phonological loop, or for verbal recall, phonological representations will be accessed. Verbal recall will require all levels o f speech output processing. Picture pointing responses will require semantic representations to be accessed from phonologically
coded material. The speech processing requirements o f some STM task paradigms will be further examined in Chapters 4 and 5.
There is, therefore, scope for a reconsideration o f verbal STM research methods and findings in the light o f the importance given to speech processing skills by a number o f influential researchers. The review that follows will focus on STM research within a developmental context.